Dag Kittlaus, the co-founder of Siri, is backing Helix, a startup using voice technology to replace CAPTCHAs and other identity verification methods. According to TechCrunch, the startup bets that voice identification can verify humans while keeping personal data off corporate servers.
As AI agents like Instinct and Muse proliferate, the internet needs a way to separate humans from bots. Helix runs a voice-identification model directly on your device, even an iPhone 13, analyzing speech to confirm you're real and estimate age, with no cloud upload. The startup closed a $6.2 million Series A led by Chicago Ventures, with Kittlaus joining as adviser and backer.
The founding trio includes Alex Oberg (ex-Ticketmaster, where he built Verified Fan), Anna Dorofiyenko (ex-Live Nation and Zillow data science), and Mike Hall (Nike and Tinder alum who coded the original matching algorithm). They are positioned to address one of tech's thorniest problems: identity verification in the age of synthetic media.
Governments worldwide are enacting age-verification laws for social media, forcing platforms and users to surrender biometric data to third-party companies. Helix's model offers an alternative, voice-as-proof-of-humanity could let platforms comply with regulation without collecting fingerprints, face scans, or government IDs. For privacy-conscious users, that's a genuine advantage.
Oberg hints the technology can detect more than just "human or bot." Helix is piloting voice analysis to flag drunkenness or distress, useful for ride-share drivers or healthcare workers but ethically fraught. Who owns that biomarker data? Can employers demand it? Will insurance companies use it against you?
The startup counts Zeel, an in-home healthcare provider, as a customer. Verification for gig workers like cab drivers is on the roadmap. Kittlaus's involvement, the co-founder who made "Hey Siri" synonymous with voice AI, signals confidence that voice is the authentication frontier. Whether regulators and privacy advocates accept it is the open question.

